Yawning:
a possible confounding variable in EMG
biofeedback
Oman RE, Sullivan SJ, Fleury J, Dutil E
Centre de Recherche,
Institut de Réadaptation de
Montréal, 6300 Avenue Darlington,
Montréal, Quebec, Canada H3S
2J4
The electrical activity recorded from a
resting or contracting muscle represents the
output of the lower motoneuron pool or final
common pathway innervating that muscle. As such,
it is reflective of a wide range of central
neurophysiological processes, both facilitatory
and inhibitory. Thus, the observed EMG signal
may reflect unrelated muscle activity in
addition to the electrical activity owing to the
variable under investigation. Such extraneous
information may include cross-talk (Wolf, 1978),
unrelated voluntary contractions (Delwaide &
Toulouse, 1981), and changes in posture (Hayes
& Sullivan, 1976). Although professionals
are aware of the peripheral spread of EMG
signals as a potential confounding variable,
they are not necessarily aware of the central,
neural, and reflexive augmentation that can
occur. This paper will demonstrate the extent of
this augmentation and that it can occur
inadvertently with a commonly occurring event,
yawning.
A recent report by Wimalaratna and Capiledo
(1988) described a case of an acute hemiplegic
patient who was unable to voluntarily contract
the muscles of the arm (a manual muscle test
score of 0 on a 5-point scale) yet was observed
to move the arm reflexively while yawning. This
observation is in keeping with that described by
Mulley (1982), who found that 31 of the 40
hemiplegic patients he questioned reported that
their plegic arm moved involuntarily during
yawning. The physiology of yawning and its
relation to respiration are not well documented
(Wimalaratna & Capiledo, 1988; Forrester,
1988), but the observation of increased muscle
activity in the paretic or plegic limb suggests
that it may have a general facilitatory effect
on central nervous system activity, although the
exact mechanisrn is yet to be determined.
A similar pattern of increased muscle
activity during yawning in a hemiplegic patient
was observed in our laboratory during an EMG
biofeedback training program. We were fortuitous
in being able to document EMG levels in the
anterior deltoid muscle both during yawning and
during a series of attempted maximum voluntary
contractions. The purpose of this paper is to
present these data and to discuss the possible
consequences of yawning in the interpretation of
data obtained in EMG biofeedback studies.
METHOD
Subject : The subject was a 71-year (right
hand dominant) male with left-sided hemiplegia
as a consequence of a cerebrovascular accident 6
months previously. The patient was not
experiencing any other health-related problems
and was not taking any medication that would
influence the outcome of the study. At the time
of the data collection he was undergoing a
comprehensive rehabilitation program and, in
addition, was participating in an optional EMG
biofeedback training program.
Prior to beginning the biofeedback training
program, his stage of motor recovery was
assessed as being at level 1 (the lowest
possible score) on the Brunnstrom scale
(Brunnstrom, 1956) and a score of 4 on a
possible of 66 on the Fugl-Meyer test
(Fugl-Meyer, Jaasko, Leyman, Olsson, &
Steglind, 1975). This score reflects the
subject's inability to dissociate synergistic
movements during an attempted voluntary
contraction of the upper limb. The patient was
unable to perform isolated movements of the
upper extremity (e.g., elbow flexion) without
contracting a number of other muscle-related
groups (e.g., shoulder flexors). Despite the
obvious plegic nature of the patient's upper
extremity (he was virtually unable to move his
arm) as evidenced by these scores, he was highly
motivated and interested in the training
program. He was, however, continually fatigued
and occasionally yawned. While he was yawning,
his "plegic" arm was often observed to move with
components of both shoulder and elbow
flexion.
Procedure : The details of the testing
procedures were presented elsewhere (Sullivan,
Oman, Mathieu, Arsenault, & Dutil, 1988) and
are presented in summary form here. Before,
immediately after, and 1 month following an EMG
biofeedback training program, the electrical
activity of the anterior deltoid was recorded
together with the motor and functional
performance of the hemiplegic/paretic limb.
During each of the 12 training sessions the
patient was seated in a chair that provided
maximum support and isolated the performance of
the shoulder flexor muscle group. He was
required to produce a maximum voluntary
isometric contraction of the shoulder flexors of
the noninvolved limb followed by the hemiplegic
limb. The electrical activity of the contracting
anterior deltoid was recorded (surface
electrodes placed 3 centimeters apart) during
each 5-second contraction. The patient was
instructed to produce a maximum voluntary
contraction and to match both the level and
pattern of the electrical activity (linear
envelope) produced by the hemiplegic limb to
that previously generated by the noninvolved
limb for each trial. The subject looked straight
ahead at the computer screen on which the
resulting feedback signals were presented. Ten
trials were recorded for each training
session.
The recorded EMG signals were amplified (CMRR
> 100 dB) and filtered (10-500 Hz) prior to
being sampled (1,000 Hz) and digitized by an IBM
PC/AT, which generated the "real-time" EMG
displays that the subject viewed. The sampled
EMG data were also trans ferred to a PDP 11 /23
+ for subsequent off-line analysis. The 5
seconds of data underwent a series of
mathematical signal processing stages as
follows. The signal was full wave rectified and
filtered at 6 Hz, and the peak (and integrated)
EMG values were obtained for each trial.
During four different training sessions EMG
data were obtained during at least part of a
yawn. These data were obtained while the patient
was resting between trials. Since yawning is an
involuntary action, and therefore not
predictable, data acquisition began when the
researcher observed the "start" of a yawn.
Consequently, data corresponding to the
beginning of the yawn were not collected. The
computer program was designed to collect 5
seconds of data from the point of initiation.
The data were not obtained in conjunction with
an attempted voluntary contraction. Synergistic
contractions of the hemiplegic limb were
observed during yawning. Owing to the restrained
nature of the subject's trunk and arm position,
the most obvious movement was that of elbow
flexion.
RESULTS
The peak values obtained from the rectified
and filtered EMG signal recorded from the plegic
limb for four 5-second intervals of yawning are
presented in Table 1. Also shown are the means
and standard deviations of peak EMG for both
limbs for the 10 trials recorded during the same
training session. For each training session
there is a marked reduction in the peak EMG
recorded from the plegic limb in comparison with
that recorded from the noninvolved limb during
an attempted maximum contraction. For each of
the four episodes of yawning the peak EMG is
markedly increased, compared with the session
means and standard deviations for the limb. This
increase is clearly beyond the normal variation
of the hemiplegic limb, and in some instances
(Trials 1 and 2) the peak EMG values for the
hemiplegic limb fall within the range of values
recorded from the noninvolved limb for that
particular session.
Sample data are shown in Figure 1. Here it
can be seen that the peak EMG value is a
realistic representation of the recorded signal,
which is indeed elevated for a large part of the
signal. Measures of the integrated EMG were not
used in this analysis owing to their lack of
comparative meaning. It should be noted that it
was impossible to record the signal in its
entirety because of the involuntary nature of
yawning.
DISCUSSION
Yawning, an involuntary motor response most
probabIy triggered from brainstem centers
(Wimalaratna & Capiledo, 1988), appears to
be capable of providing a central drive or
facilitation to peripheral motoneurons in the
hemiplegic patient. We documented a marked
increase in only a single muscle; however,
coupled with our clinical observations of the
development of a general synergistic flexion of
the plegic limb, this clearly suggests a central
facilitation of the otherwise flaccid limb.
The changes observed here support previous
observations (Mulley, 1982; Wimalaratna &
Capiledo, 1988) and also document the magnitude
of the facilitation at the level of the
individual muscle. The facilitation allowed the
muscle to obtain an output level approximate to
that recorded from the homologous muscle of the
noninvolved side. This occurred for two of the
four episodes of yawning. The other two episodes
also resulted in greatly increased electrical
output. This facilitation may be likened to the
changes observed with the use of
facilitatory/inhibitory techniques by physical
therapists in the treatment of CNS disorders
(Bobath, 1978). The use of such techniques
requires the patient to either voluntariIy or
passively adopt a specified posture in order to
provide the desired facilitation/inhibition.
Obviously, an involuntary phenomenon such as
yawning cannot be used in the same manner. The
observations in this case clearly suggest that
the peripheral motoneurons are capable of being
activated if the appropriate central drive is
provided.
In EMG biofeedback studies, changes in the
level of electrical activity recorded from the
contracting muscle are used both for feedback
purposes and for the direct quantification of
the muscle activity. Fundamental to the use of
the EMG signal is the fidelity of the signal at
both the technical (Cram & Garber, 1986;
Sullivan, Mathieu, & Goulet, 1989) and
physiological levels (Basmajian, 1985). The
observations reported here clearly suggest that
the
EMG signal could be contaminated by yawning,
thus leading to possible misinterpretation of a
particular value. A related concern is the
possibility of greatly increased variability
over a series of trials. Just one episode of
yawning would inflate the variability of a
training session, leading to difficulty in
interpreting the data. The data presented in
Table 1 clearly show the degree of variability
associated with the recording of peak EMG values
in a controlled laboratory situation. Peak EMG
recordings that represent the single greatest
value recorded during the contraction are
inherently variable both within and between
recording sessions. This, coupled with the
changes due to training (e.g., trials 1 to 4
represent a difference of 12 training sessions)
may account for changes in both absolute values
and the associated variability observed in the
present data. In a clinical situation, there may
very well be a further increase in variability,
and inclusion of data due to yawning or related
involuntary activity would further add to this
situation. Possible misinterpretations are most
likely to occur in situations where the data
acquisition system is highly automated and the
results of individual trials are not routinely
monitored for EMG artifacts. This situation is
of paramount importance when working with
hemiplegic patients who often appear fatigued
and consequently have the potential for episodes
of yawning. The potential for yawning's being a
confounding variable in EMG bio feedback studies
must be given serious consideration in both
clinical and research studies. Although the
exact physiological mechanism underlying yawning
is yet to be established, it does appear to
exert a strong facilitatory influence on the
level of muscle activity generated by the
hemiplegic patient.
REFERENCES
Basmajian, J. V. (l 985). Muscles alive.-
Their functions revealed by electromyography
(5th ed.). Baltimore: Williams &
Wilkins.
Bobath, B. (1978). Aduit hemiplegia:
Evaluation and treatment (2nd ed.). London:
Heinemann Medical Books.
Brunnstrom, S. (1956). Associated reactions
of the upper extremity in adult patients with
hemiplegia. Physical Therapy Review, 36(4),
225-236.
Cram, J. R., & Garber, A. (1986). The
relationship between narrow and wide band-width
filter settings during an EMG scanning
procedure. Biofeedback and Self-Regulation, 11,
105-114.
Delwaide, P. J., & Toulouse, P. (l 98
1). Facilitation of monosynaptic reflexes by
voluntary contractions of muscles in remote
parts of the body: Mechanisms involved in the
Jendrassik manoeuvre. Brain, 104, 701-719.
Forrester, J.
M. (1988). Is yawning a brainstem
phenomenon? Lancer, March, 596.
Fugl-Meyer, A. R., Jaasko, L., Leyman, I.,
Olsson, S., & Steglind, S. (1975). The
post-stroke hemiplegic patient: A method for
evaluation of physical performance. Scandinavian
Journal of Rehabilitation Medicine, 7,
13-33.
Hayes, K. C., & Sullivan, S. J. (1976).
Tonic neck reflex influence on tendon and
Hoffmann reflexes in man. Electromyography and
Clinical Neurophysiology, 16, 250-261.
Mulley, G.
(1982). Associated reactions in the hemiplegic
arm. Scandinavian Journal of Rehabilitation
Medicine, 14, 177-120.
Sullivan, S. J., Mathieu, P. A., &
Goulet, C. (1989). The dilemma of EMG filter
bandwidth selection for biofeedback studies -A
deterministic investigation. Part 1: RMS
analysis. Proceedingsof the2Oth AnnualMeetingof
the Association for Applied Psychophysiology and
Biofeedback (pp. 215-218). San Diego,
California: Association for Applied
Psychophysiology and Biofeedback.
Sullivan, S. J., Oman, R. E., Mathieu, P.
A., Arsenault, A. B., & Dutil, E. (1988). A
template model, microcomputer based EMG
biofeedback system for use with hemiplegic
patients. Proceedings of the International
Conférence of the Association for the
Advancement of Rehabilitation Technology (pp.
430-431). Montreal, Quebec, Canada: RESNA.
Wimalaratna, H.
S. K., & Capiledo, R. (1988). Is yawning
a brainstem phenomenon? Lancet, February,
300.
Wolf, S. L. (1978). Essential considerations
in the use of EMG biofeedback. Physical Therapy,
58, 25-31.
Is yawning a
brainstem phenomenon ? Wimalaratana HS,
Capildeo R. A stroke patient who stretched his
hemiplegic arm during yawning
Lancet 1988; 1; 8580;
300
